Printer Friendly

The "passive" voice in Japanese and German: argument reduction versus argument extension (1).


In this article, we will examine the similarities and differences between the passive forms of German and Japanese. We would like to argue that the structural similarity observed in the passive forms is the result of the deprofiling of the agent, and that the most salient difference between both passives, namely the existence of the Japanese adversative passive, can be explained in line with a general property of the Japanese language, which is the ability to extend its argument structure by adding the experiencer as an extra-argument.

Through the discussion above, the "traditional" definition of passive (that a marked verbal morphology and a ease alternation describe an event centripetally from the viewpoint of the patient) will be modified to a more universal treatment of the voice category. This universal treatment is compatible with a higher degree of variation among languages.

1. Introduction

Many languages have the capacity to encode the same propositional values with more than one case framework and verbal morphology. Both in Japanese and German, we have active sentences and their counterparts which are labeled passive or "Ukemi"-forms. In this article, we will examine the similarities and differences between the passive forms of German and Japanese and propose:

1. that there is a structural similarity between both "passive" forms in German and Japanese (i.e. reduction of an argument, agent), which results in a functional similarity (reflecting the hierarchy of transitivity), and

2. that the specific function of the Japanese adversative passive can be explained in line with a general property of the Japanese language (i.e. extension of an extra-argument, experiencer).

We have two major reasons for choosing these two languages to analyze the phenomena of passive: the most straightforward reason being that we are Japanese philologists of the German language who have the greatest knowledge of these two languages among other languages, and the second, more linguistic reason is that the German language allows for passive sentences with intransitive verbs and provides informative data for illustrating the role of transitivity in passive formation, and that the Japanese language can also construct passive sentences from intransitive ones without agents.

In German, the passive is opposed to the active and both of them taken together form the diathesis paradigm. The Japanese passive, the (r)are-construction, (2) cannot easily be located in the diathesis system. This difficulty arises from the definition of the diathesis itself; in European languages the active and passive voices differ from each other in the following ways:

(1) "Traditional" definition of the passive (cf. Muraki 1991)

a. Verb morphology: the verb obtains a more marked form in passive than in active clauses. In German the eventive passive (Vorgangspassiv) consists of a past participle and the auxiliary verb werden 'become'.

b. Syntactic-semantic structure of the sentence: in the active voice, the subject is usually the agent, but it is the patient in the passive voice. The action is centrifugal in active phrases; that is, it is carried out by the subject, whereas it is centripetal in passive phrases in that it is directed at the subject.

c. Case marking: in active sentences, agents are normally coded in the nominative case and patients in the accusative case. In passive forms, patients are coded in the nominative case and subjects are either suppressed or occur in an oblique case.

In (1), we have a standard definition of the passive. The Japanese (r)are-construction seems to satisfy all of these criteria and might be treated as passive; it contains the marked morphology and codes the patient in the ga-case (nominative), and the subjects are either suppressed or are coded in the ni- or kara-case (oblique case). We must, however, note three specific characteristics of the (r)are-construction. Firstly, the auxiliary (r)are is used not only for the formation of passive but also in cases of spontaneity, possibility, and honorific forms. This suggests that the (r)are-construction has the function of reducing transitivity and that the passive meaning would result from it. Secondly, the criteria of diathesis above are not just met by the (r)are-construction but also by causative, reflexive, reciprocal, and yaru-kureru-sentences (cf. Muraki 1991). Thirdly, there is a peculiar usage of the passive in Japanese, the so-called adversative passive, which would not be classified as passive in the above definition. We will further address this problem by discussing the features of German and Japanese passive forms, which will lead us to a new concept of the voice paradigm.

2. The German passive

German passive has an agency restriction. Verbs which cannot have an agent subject cannot be used to construct passive sentences.

(2) *Ein schoner Sommer wird von uns in diesem Jahr gehabt.

a nice summer PASS by us in this year had

(Eroms 2000: 400)

This restriction holds for the intransitive passive as well.

(3) *Ihm wurde von dem Unternehmen gelungen.

him.DAT PASS.PAST by the enterprise succeeded

This restriction is, however, not an absolute one. Passive sentences with nonagentive verbs can be found sporadically. Consider (4) and (5).

(4) Hier wird geblieben!

here PASS stayed

'You must stay here!'

(Zifonun et al. 1997: 1805)

(5) In den Lazaretten wird gelitten und gestorben.

in the military.hospitals PASS suffered and died

'There are many who are suffering and dying in the military


(Eroms 2000: 428)

These examples show that agency is not an absolute restriction, but a preference rule. We can assume, therefore, that there is no strict binary rule for passive usage; there is only a continuous one: what we expect in the usage of the German passive is not a general prohibition of verbs with non-agent subjects, but a gradual preference of verbs for the passive.

A corpus analysis carried out with the Mannheimer Korpus confirms this assumption; some verbs tend to be used in the passive, while others do not. The principle that determines this affinity is transitivity. We used the parameters of transitivity in Hopper and Thompson (1980) to measure the transitivity degree.

Verbs with many properties from the middle column in Table 1 are highly transitive. Therefore, a prototypical transitive verb has two or more participants; describes a telic, punctual, and volitional action; and is an affirmative, indicative sentence with an individuated, affected object.

Based on this scale of transitivity, the results from the Mannheimer Korpus were as follows.

In Table 2, we can see that the passive form appears more often with highly transitive verbs. Here the verbs are arranged according to their transitivity. Verbs describing a state change like ermorden 'murder' and toten 'kill' have the highest degree of transitivity, and those describing a physical activity like brechen 'break' and schlagen 'hit' have a higher transitivity value than those describing a cognitive activity like finden 'find' or wunsehen 'wish'. Verbs like feiern 'celebrate' can be used both transitively and intransitively.

We can see clearly that the passive affinity corresponds to the transitivity degree: verbs of high transitivity such as ermorden or stehlen are most frequently used (approx. 40%) in the passive voice. On the contrary, we rarely find the passive with verbs like passive with bleiben 'stay'. Phrases in which the passive voice is used with intransitive verbs are infrequent, but they do occur.

The corpus analysis shows quantitative differences depending on the degree of transitivity. We will now address the meaning of the passive voice and attempt to answer the question of whether or not there are any qualitative differences according to verbal transitivity.

Weinrich (1993) pointed out that the intransitive passive, (3) a nontypical passive form in German, has a salient property in its usage in a text: the intransitive passive is highly dependent upon its situation. Since an intransitive passive has no formal subject, the hearer depends heavily on supplementary information of the surrounding situation. It follows, then, that the intransitive passive should occur more frequently in the imperative (6) or the interrogative (7).

(6) Jetzt wird endlich mal gearbeitet!

now PASS at.last for.once worked

'Will you finally get to work now!'

(Weinrich 1993: 177)

(7) Wann wird sich endlich mal gekammt?

when eASS REFL at.last for.once combed

'When will you finally comb your hair?'

(Weinrich 1993: 177)

Our corpus analysis confirms this hypothesis. The intransitive passive shows an affinity for an illocutionarily marked environment: (4) the passive with low-transitive verbs, typically the intransitive passive, is often used in an order or a question. On the contrary, the passive prototype, the transitive passive, does not show such a tendency. Verbs with lower transitivity tend to be used with marked illocution, while the impersonal or interrogative usage of the prototypical transitive verbs is not salient.

Besides the saliency of its illocution, there is a further difference in the passive function. Let us have a closer look at the indicative usage of the passive. By the indicative usage, we mean a passive which describes a real state of affairs. Namely, the indicative passive describes an event from the reverse perspective, which is a prototypical function of the passive construction.

A closer look shows that the indicative use can be classified further into at least three subclasses. In some cases, the passive describes a concrete, individual action. In other cases, the passive does not describe an individual action, but an iterative action or a habitual event in the past, which is not regarded as individual actions, but perceived as a coherent whole. Table 3 shows how the indicative passive is used.

Here, we can observe a clear tendency: the lower the transitivity of a verb, the more frequently we find cases in which the passive expresses an iterative/habitual event. The passive of toten 'kill' almost always depicts a singular action, while habitual actions are construed as passives in about 60% of the occurrences of passive arbeiten 'work'.

To summarize, the agent rule for verbs that can be used in the passive is a rule of tendency. The transitivity of verbs plays an important role in quantitative as well as qualitative aspects of the passive usage. A verb with high transitivity shows a passive affinity, and in this case the passive is a means to express an action from another point of view. The passive with low transitive verbs is not a prototypical one, and it has a certain marked-illocutionary coloring, or it describes an iterative/habitual process. At the extreme end of low transitivity are unaccusative verbs. They are seldom used in the passive voice; the passive has a strongly marked meaning. Figure 1 depicts a summary of the German passive usage.


3. The Japanese passive (Ukemi)

The Japanese passive is composed of a verb and an auxiliary verb (r)are. This auxiliary verb (r)are is not exclusively used for passive formation, but for other functions as well. Traditionally, it is claimed that the (r)are-construction has four functions in Japanese Grammar. (cf. Karashima 1993).

(8) Functions of (r)are-construction

a. Passive

Minna-ni shokuji-wo tabe-rare-ta.

everyone-DAT meal-Ace eat-PASS-PAST

'Regrettably, the meal was eaten up by everyone.'

b. Potentiality

Kono kusa-wa tabe-rare-ru.

this grass-top eat-potential-inflection

'This grass is eatable.'

c. Honorific

Kono hon-no koto-wo

this book-gen thing-acc



'Did you remember this book?'

d. Spontaneity

kore-wo miru-to, ano hito-ga

this-ACC see-if the person-NOM



'If I see it, the person will come to my mind.'

These four usages of (r)are have been widely discussed and to some extent there is a consensus about their development. Karashima (1993) shows that the honorific use of (ra)ru (= (r)are) did not exist at the Joodai stage (in the seventh-ninth century AD) and it arose later than the other three uses. The potential use could only be found in negative sentences so that this use can be assumed to have developed later than the spontaneous and the passive uses (Shibatani 2000, cf. also Hashimoto 1969; Kuginuki 1991). Although it is widely accepted that spontaneous and passive uses are the oldest ones, there is no evidence to prove which of these uses came first. In this article, we agree with Shibatani (2000), who stipulates that the passive use arose from the spontaneous use through the conceptualization of the volitional agent as peripheral participant instead of the nonvolitional agent which is demoted from the subject position in a sentence in spontaneous use. We will, however, not discuss this problem further because the derivation of these two uses is not relevant for our discussion of the passive. We only argue here that the defocusing of the agent and the reduction of transitivity are related to all these four uses of (r)are. The spontaneous use of (r)are leads to the construal of an event as a quasi-automatic process without mentioning its causer. In the honorific usage we refrain from referring to the actor, while the agentivity of the actor is backgrounded entirely. In the usage of potentiality, the process in question will be described as "possible to occur"; the ability of the actor will not be expressed by (r)are, but rather affectability of the object. The agent is left out of the picture.

Now let us have a closer look at the passive usage of the (r)are-construction. The Japanese passive can be subdivided into two categories: the direct passive and the indirect one. We will sketch out the properties of these two passive forms.

3.1. The direct passive

(9) a. Taro-ga Jiro-ni nagur-are-ta.

Taro-NOM Jiro-by hit-PASS-PAST

'Taro was hit by Jiro.'

b. Jiro-ga Taro-wo nagu-tta.

Jiro-NOM Taro-ACC hit-PAST

'Jiro hit Taro.'

The object in the active sentence is raised to the subject in the passive. The subject in the active is realized in the ni-case in the passive. It can be suppressed. Here, the passive carries out the communicative function of focusing on the patient. In the examples above, the selection of one form is made only in terms of the intention of the speaker, namely, whether he will make the hitting person or the person who is hit the focus. In many cases, however, the selection results from the general properties of the Japanese language. There are two decisive factors here: Silverstein's (1976) hierarchy of nominal phrases and the transitivity of verbs. Japanese shows a tendency to assign the subject function to nominals that rank high on the animacy hierarchy. Therefore, each a-sentence in (10) and (11) is very unnatural, but the b-sentences are used frequently (cf. Tsunoda 1991: 45).

(10) a. ??Kuma-wa watashi-ni osow-are-ta.

bear-TOP I-by attack-PASS-PAST

'The bear was attacked by me.'

b. Watashi-wa kuma-wo oso-tta.

I-TOP bear-ACC attack-PAST

'I attacked the bear.'

(11) a. ??Kuma-wa watashi-wo oso-tta.

bear-TOP I-ACC attack-PAST

'The bear attacked me.'

b. Watashi-wa kuma-ni osow-are-ta.

I-TOP bear-by attack-PASS-PAST

'I was attacked by the bear.'

The first person (watashi) takes the superior position in the nominal hierarchy and is predisposed to be coded as the subject. For this reason, the active is chosen if the first person is the agent and the passive if it is the patient in the sentence.

Furthermore, the transitivity of a verb also affects the realization of the active and the passive forms. Generally, verbs with high transitivity are used more frequently in the passive than those with low transitivity.

(12) a. Hanako-wa Taro-wo koroshi-ta.

Hanako-TOP Taro-ACC kill-PAST

'Hanako killed Taro.'

b. Taro-wa Hanako-ni koros-are-ta.

Taro-TOP Hanako-by kill-PASS-PAST

'Taro was killed by Hanako.'

(13) a. Hanako-wa Taro-wo nagu-tta.

Hanako-TOP Taro-ACC hit-PAST

'Hanako hit Taro.'

b. Taro-wa Hanako-ni nagur-are-ta.

Taro-TOP Hanako-by hit-PASS-PAST

'Taro was hit by Hanako.'

(14) a. Hanako-wa Taro-wo ma-tta.

Hanako-TOP Taro-Ace wait-PAST

'Hanako waited for Taro.'

b. ?Taro-wa Hanako-ni mat-are-ta.

Taro-TOP Hanako-by wait-PASS-PAST

'Taro was waited for by Hanako.'

(15) a. Hanako-wa shukudai-wo omoidashi-ta.

Hanako-TOP homework-ACC remember-PAST

'Hanako remembered her homework.'

b. * Shukudai-wa Hanako-ni omoidas-are-ta.

homework-TOP Hanako-by remember-PASS-PAST

intended: 'Her homework was remembered by Hanako.'

(16) a. Hanako-wa sono hon-wo motteiru.

Hanako-TOP the book-ACC have

'Hanako has the book.'

b. * Sono hon-wa Hanako-ni mot-are-teiru.

Taro-TOP Hanako-by have-PASS-INFLECTION

'* The book is had by Hanako.'

The relation between the transitivity of verbs and the acceptability of passive sentences can be depicted as in Figure 2 below. With regard to transitivity, the Japanese passive shows a distribution of verbs similar to what we observed in the German passive above.


3.2. The indirect passive

In the direct passive, the argument that codes the person affected by the event is chosen as the subject of the sentence. This argument, however, does not exist in the argument structure of the indirect verb. In this way, the indirect passive is distinguished from the direct passive, which has either the same number of arguments as active or a smaller number of arguments caused by the elimination of the agent.

(17) a. Ame-ga fu-tta.

rain-NOM fall-PAST

'It was raining.'

b. Watashi-wa ame-ni fur-are-ta.

I-TOP rain-by fall-PASS-PAST

'I got caught in the rain.'

(18) a. Musuko-ga shin-da.

son-NOM die-PAST

'The son died.'

b. Sono hahaoya-wa musuko-nishin-are-ta.

the mother-TOP son-by die-PASS-PAST

'Unfortunately for the mother, the son died.'

These sentences show that the "experiencer" suffers damage through the occurrence of an event. Therefore, they are often called adversative passive sentences. Indirect passive sentences may relate to transitive active counterparts just as well as to intransitive active counterparts.

(19) a. Watashi-wa kare-ni kao-wo nagur-are-ta.

I-TOP he-by face-ACC hit-PASS-PAST

'I was hit in the face by him.'

b. Kare-wa watashi-no kao-wo nagu-tta.

he-TOP I-GEN face-ACC hit-PAST

'He hit me in the face.'

The distinguishing feature of the sentence (19a) is the presence of an accusative object. The available research states that this kind of sentence occupies an intermediate position between direct and indirect passive sentences (cf. Nitta 1991). However, we regard them as a subclass of indirect passive sentences. In order to address this matter, we need to consider the semantic properties of the accusative objects in passive sentences.

(20) Kanojo-wa oshiri-wo sawar-are-ta.

she-TOP back-ACC touch-PASS-PAST

'She was touched on the back.'

(21) Watashi-wa fuini kata-wo tatak-are-ta.

I-TOP suddenly shoulder-ACC pat-PASS-PAST

'I was suddenly patted on the shoulder.'

The acceptability of the sentences will clearly decline if the accusative denotes an alienable possessum, especially when the verb is not highly transitive.

(22) a. Watashi-wa computer-wo kowas-are-ta.

--TOP computer-ACC destroy-PASS-PAST

'I was adversely affected by someone's destroying my computer.'

b. ??Watashi-wa computer-wo sawar-are-ta.

.--TOP computer--touch-PASS-PAST

'I was adversely affected by someone's touching my computer.'

Given that high transitivity is required to form indirect passive sentences with an accusative object, this form shows similarities with the direct passive. On the other hand, it has a common property with the indirect passive without an accusative object because the experiencer is negatively affected by the event. We can now point out the possibility that these three forms could be placed on one continuous scale. Before continuing with this argument, we would like to characterize the German passive as a form of argument reduction by clarifying the similarities and differences between the two types of passive forms.

4. Commonness: argument reduction

We can assume that both passive constructions share a common function. In both Japanese and German, the passive suppresses the first argument, the agent. Further, we observed in both languages that an event of high transitivity is more likely to be coded in the passive: prototypically, a passive is used for a perfective, punctual, and intended action with a direct object. The action is real, carried out by a causer, and changes the status of the object.

These syntactic-semantic characteristics of the passive yield similar restrictions or distributions on the one hand. On the other hand, there are certain divergences between the two constructions. We will now try to describe how these similarities and differences of the passive function relate to each other.

4.1. Prototype of the passive: description of an action from a reversed perspective

In Section 1, we observed that prototypical transitive verbs are also prototypical passive verbs. A passive with the prototypical transitive verb expresses a real event from the non-agent point of view. A "good" passive in German is formed with a transitive verb which represents a perfective, intended activity with an effected object.

German and Japanese passives are similar in terms of distribution with regard to transitivity. A highly transitive verb allows a passive form without any particular conditions, while a low transitive verb goes along with semantic restrictions on forming the passive. Verbs with positive values in the transitivity parameters in Table 1 are verbs from which the passive is easily formed. These verbs describe a strongly affecting action both in the active voice and the passive voice.

4.2. Two divergent directions from the prototype

The lower the transitivity of a verb is, the less frequently the verb is used in the passive. A nonprototypical passive has a nonprototypical meaning: in this nonprototypical usage, the passive is no more than a syntactic form for expressing an action. In this, we can observe two divergent directions from the passive prototype: the passive is used either in an illocutionarily marked context or in the environment where a perfective event is interpreted as an imperfective one.

4.2.1. Passive as a modal expression. We have seen that the intransitive passive is often illocutionarily marked. It is used to give an order, formulate a wish, or to ask a question.

(23) Jetzt wird fleissig gearbeitet!

now PASS diligently worked

'You must work hard now!'

(24) Jetzt wird aber endlich mit dem Telefonieren

now PASS but finally with the telephoning

Schluss gemacht!


'You must stop talking on the phone right now!'

(Weinrich 1993: 397)

That is a striking contrast to the Japanese (r)are-honorific. As we saw above, the (r)are-construction is used not only in the passive, but also as an honorific form.

(25) Nichiyoubi-mo hatarak-are-ru node ...

Sunday-also work-RARE-INFLECTION because

passive: 'Because he, to my regret, works even on Sunday ...'

honorific: 'Because he, a person of good standing, works even on Sunday ...'

(25) can be interpreted either as a passive or as an honorific. The actual context determines the meaning of (25). The Japanese passive form has honorific meaning and the German passive gives a curt order. It is contradictory but formed in a related manner. In both of the passive forms, there is no agent of action. In German, the absence of the agent results in a harsh order which allows no objection. In Japanese, the speaker avoids naming the agent of action because it is inappropriate to mention a person in a high position. The same characteristic, namely, the absence of the agent, turns out to have opposite meanings.

As Weinrich (1993) pointed out, the peculiarity of the intransitive passive can be traced back to its situational dependency. Without a formal subject, which could guarantee a close relation to the context in a prototypical passive, the intransitive passive requires another binding to the situation. Whether it renders an order or a question, the intransitive passive is embedded in context. There are also contracted forms of the intransitive passive.

(26) Aufgepasst! (< Jetzt wird [aber] aufgepasst!)

paid.attention now PASS but paid.attention

'Pay attention!'

4.2.2. Iterativity/habituality. Another aspect of the passive with low transitive verbs is iterativity/habituality. In the iterative/habitual usage, it does not represent an individual, concrete process, but a collective, reiterating process. The typical example is the passive with an unaccusative verb.

(27) [Frankfurt's central station is an important junction in Europe.]

Hier wird bei Tag und Nacht angekommen und

here PASS by day and night arrived and abgefahren.


'People arrive here and depart from here day and night.'

An unaccusative verb is rarely used in the passive, but, without an agent, the unaccusative verb can be used in the passive if the passive expresses an iterative process. In this case, a perfective event is interpreted as a continuous, imperfective event. Here we also observe that the transitive parameters, namely telicity, punctuality, and individuality, turn to the negative value and the described event is no more thought of as an action, but as a pure process.

In the Japanese passive, this effect of an iterative meaning with certain passives is also observed.

(28) a. * Kono uta-wa, Seino-san-ni utaw-are-ru.

this song-TOP Seino-Mr.-by sing-PASS-INFLECTION

'This song is sung by Mr. Seino.'

b. Kono uta-wa yoku Seino-san-ni utaw-are-ru.

this song-TOP often Seino-Mr.-by sing-PASS-INFLECTION

'This song is often sung by Mr. Seino.'

The odd passive with an inanimate subject as in (28a) will become much more grammatical if it is not understood as a concrete, singular action, but rather as an iterative event as in (28b).

4.2.3. Indirect passive--passive of suffering. The Japanese indirect passive has another specific reading. This reading cannot be handled within the parameter frame mentioned above. It is likely that transitivity plays little part here. A detailed analysis will follow in Section 5.

4.3. Passive form and the category of the "passive"

The similarities between the German and the Japanese passives are not only functional, but morphological as well: in both languages, passives feature an existence verb and an infinite verb.

In German, the auxiliary verb werden is the agent absorber that depicts an event without mentioning the agent. Werden, with its mutative meaning, has become more common than the copula verb sein, which played a more important role in a former stage of the language (cf. Eroms 1990; Kotin 1995).

The Japanese passive is built with a verb and the auxiliary verb (r)are. This combination is, as we saw above, not just reserved for the passive, but appears in forms of spontaneity, possibility, and the honorific as well. These usages of the (r)are-construction have a common property: to avoid stating the agent explicitly. In the spontaneity usage, the event occurs quasi-automatically, and in the possibility usage, the possibility in question is affectability of the object. In the honorific usage, the event is described as if the subject had no influence on the process. In the passive, this agent-deprofiling function is exploited to convey a typical passive function.

The morphology of the passive expresses the suppression of the agent. In both languages, the absence of the agent in the passive with a highly transitive verb is interpreted as an alternative description of an action. The low transitive verbs get a special meaning specific to the language. Here, we can see a possibility of maintaining the common category "passive."

4.4. Summary of the similarities

An observation from distributional and morphological properties of the German and the Japanese passive forms yields the following frame for the category "passive."

In both passive forms, the agent is deprofiled. This creates the basis for the meaning of the passive. The passive prototype is built with a highly transitive verb which describes an action from an alternative point of view. The deprofiling of the agent leads to apparently different but, at closer look, similar usages in Japanese and German. In German, the illocutionary markedness or the iterativity/habituality gets profiled. In Japanese, we have another specific use of (r)are-verb forms with the indirect passive whose function seems to extend beyond the frame of the transitivity parameters.

5. Differences: argument extension

Now we will turn to the differences between both constructions. The Japanese passive has a peculiar usage, the indirect passive, which has the contrastive property of extending the argument frame of a predication/ verb.

In Section 2, we pointed out the possibility of regarding the direct passive, the indirect passive with an accusative object and the indirect passive without an accusative object, as points on a continuum. Consider the following examples:

(29) a. Kare-wa nagur-are-ta.


'He was hit.'

b. Kare-wa kao-wo nagur-are-ta.

he-TOP face-ACC hit-PASS-PAST

'He was hit in the face.'

These sentences express only one event, because naguru means that one hits someone's face. The sentence (29b) only specifies which part of the body was hit. We can also find very similar examples in German.

(30) a. Er wurde geschlagen.

he PASS hit

'He was hit.'

b. Er wurde ins Gesicht geschlagen.

he PASS into.the face hit

'He was hit in the face.'

Such valence extension is only possible by using prepositional phrases in German, whereas in Japanese it can occur by adding an accusative object. Therefore, there is good reason to treat the direct and indirect passive forms as two figures of one and the same category. It is actually possible if we assume the existence of a pro-object jibun 'oneself' in the direct passive, as can be seen from the examples below.

(31) a. Kanojo-wa kare-wo nagu-tta.

she-TOP he-ACC hit-PAST

'She hit him.'

b. Kare-wa kanojo-ni [phi] nagur-are-ta.

he-TOP she-by hit-PASS-PAST

'He was hit by her.'

c. Kare-wa kanojo-ni jibun-wo nagur-are-ta.

he-TOP she-by REFL-ACC hit-PASS-PAST

'He was adversely affected by her hitting him.'

d. Kare-wa kanojo-ni musuko-wonagur-are-ta.

he-TOP she-by son-ACC hit-PASS-PAST

'He was adversely affected by her hitting his son.'

Sentence (31b) is not derived from (31a), but by way of obligatory deletion of jibun-wo from (31c). (31c) sounds artificial. It is well-formed if it is used in contrast with sentence (31d).

(32) Kare-wa musuko-wonagur-are-ta node-wa naku,

he-TOP son-ACC hit-PASS-PAST because-TOP not

jibun-wo nagur-are-tano da.

REFL-ACC hit-PASS-PAST because is

'He was adversely affected not because his son was beaten, but because he himself was beaten.'

The fact that the direct passive is also possible without expressing the damage of the affected person can be explained as follows. Consider (33):

(33) Seito-wa sensei-ni home-rare-ta.

student-TOP teacher-by commend-PASS-PAST

'The student was commended by the teacher.'

The fact that (33) expresses the profit of the student provides a good reason for the direct passive to be distinguished from the indirect passive. However, we can also assume the same construction as (31c) here.

(34) Seito-wa [sensei-ni jibun-wo] home-rare-ta. student-TOP teacher-by REFL-ACC commend-PASS-PAST 'The student was commended about himself by the teacher.'

The reference of the accusative object in the event is always identical with the affected person. Jibun stands logically in closer proximity to the possessor than the inalienable possessum because seito and jibun in (34) have identical referents. If an argument in the event is identical with the affected person, it can be realized as an external experiencer that is added to the proposition, even if the influence is of positive nature.

From these observations we can now bring together all three passive forms in Japanese to one basic formula and characterize it as follows.

(35) Y-wa/ga X-ni Z-wo V-(r)are-ru


i. If Z and Y corefer, that is, if Z is jibun, it must be deleted, and then the direct passive is generated. As a prerequisite for this, the verb must have a certain nonminimal degree of transitivity so that it is capable of expressing an effect on Y.

ii. If Z is inalienably possessed (e.g. as a body part) by Y, it can easily be realized in the accusative, because we can imagine that X's action affects Y immediately. If Z is not inalienably possessed by Y, the extent to which Y is affected must be estimated high enough that the indirect passive is possible. The transitivity must be correspondingly high. The influence is usually of negative nature.

iii. If the verb is intransitive, Z will not be realized because it does not exist in the argument structure of the verb. The restriction of verbs is not as strong as in case (ii) above because the event concerns Y directly and the affectedness of Y is easily conceivable. In this case, we always find the adversative passive.

6. Conclusion

We have seen what both the German passive and the Japanese passive have in common, and how they differ from each other.

The common property we observed is that the agent is deprofiled in both languages. Prototypically, this passive property results in changing the perspectives. If the agent is not salient enough to be suppressed, the passive gets one of the more specific meanings discussed above.

The most salient difference between the German passive and the Japanese passive is their respective argument taking properties. In German, the passive formation leads to a valence reduction of one argument, viz. the agent, as is often the case in passive constructions, while the Japanese passive extends its argument structure by one argument, viz. the experiencer.

How can we coherently explain this contradicting observation? Are the passive phenomena in German and Japanese to be subsumed in the same category? Or do they belong to different categories, despite the fact that there are some common functions in both constructions? From the discussion so far, we argue that the common aspect of the passive, the deprofiling of the agent, is the fundamental function of the passive and the difference is a result of a language specific property of Japanese.

It is well known that there are some other cases of the argument extension in Japanese besides that of the passive. The double subject construction, as we call it, is an example of such an argument extension.

(36) Zou-wa, hana-ga nagai.

elephant-TOP nose-NOM long

'As for elephants, they have long trunks.'

In this classical example of a so-called double subject sentence, we find the same constellation of argument binding. Between the one (the elephant) and other (the nose), there must be a part-whole relation as was the case in (35).

In Japanese, this argument extension is an important means of predication. Felix (2003: 131) argues that "the theta criterion, [a theoretical hypothesis which represents the argument structure of the sentence], does not hold in Japanese and theta roles are determined contextually with argument NPs being optional rather than obligatory." That is, an argument which is added to the argument structure as an extra-argument is licensed contextually.

Tanaka (2003) suggests that this canonical principle of the sentence structure, theta criterion, can be canceled when generating the topic construction: that is, two arguments of the same theta role can appear in one sentence or an extra-argument-topic which is not selected by the predicate can be added to the sentence. We could argue that the argument extension in the Japanese passive sentence is a case of this topic constraint. As a matter of fact, the first argument of the Japanese passive, that of the experiencer, is almost always coded as a topic. (5)

Adding an extra-argument is not an idiosyncratic property of the Japanese language. Hole (this volume) proposes a general rule to the licensing of extra-arguments:

(37) The extra argument "binds" the unsaturated variable of the c-commanded relational noun within the DP that denotes the possessum or (body) part of the referent of the extra argument. (Hole this volume) (6)

Actually, the Japanese indirect passive seems to follow this general rule for extra-argumentality.

(38) Kanojo-wa, musuko-nishin-are-ta.

she-TOP son-by die-PASS-PAST

'She was adversely affected by her son dying.' ('mother' and 'son' as a possessor-possessum relation)

According to Hole (this volume), the semantic roles added are not canonical ones like agent or patient, but "oblique" ones like experiencer or beneficiary. The semantic role which is added in the Japanese indirect passive is also either a beneficiary or a maleficiary, as we saw above.

On the other side, we can observe apparent exceptions to Hole's generalization with examples like (40).

(39) Hiroshima-wa kaki-ga oishii.

Hiroshima-TOP oyster-NOM delicious

'As regards Hiroshima, its oysters/the oysters there are delicious.'

(40) kaki-wa Hiroshima-ga oishii.

oyster-TOP Hiroshima-NOM delicious

'As regards oysters, Hiroshima has delicious ones.'

While there is supposed to be some sort of possessive relation between Hiroshima and kaki 'oyster' in (39), there is no such relation between the two arguments (kaki and Hiroshima) in (40). The relation is rather one between the argument kaki and an attributive proposition like 'something is delicious in Hiroshima'.

This type of argument extension, extra-argument-addition through the argument-proposition-relation, comes under observation in the Japanese indirect passive.

(41) boku-wa, ame-ni fura-re-ta.

I-TOP rain-by fall-PASS-PAST

'I was adversely affected by the rain falling.'

We can summarize our observation about extra-argument-licensing: generally in Japanese, an argument is licensed under the following conditions:

i. if a noun is a direct participant of the action of the predicate (theta criterion), or

ii. if a noun is an indirect participant of the action through the relation with a more deeply embedded argument (Hole this volume), or

iii. if the following proposition can be regarded as attribution to the noun in question.

These properties reflect the affectedness of the added argument. The less the argument is affected by the action of the predicate, the clearer an "extra-meaning," that is, the meaning of 'maleficiary' or 'beneficiary', is yielded.

(42) Taro-wa, gaikokude sodate-rare-ta.

Taro-TOP abroad bring.up-PASS-PAST

'Taro was brought up abroad.'

(43) Taro-wa, gaikokude jibun-no musume-wo sodate-rare-ta.

Taro-TOP abroad REFL-GEN daughter-Ace bring.up-PASS-PAST.

'As regards Taro, his daughter was brought up abroad.'/

'Taro was adversely affected by the fact that his daughter was brought up abroad.'/

'Taro's daughter was brought up abroad against his will.'

(44) Taro-wa, Hanako-ni michi-de mat-are-ta.

Taro-TOP Hanako-by way-LOC ambush-PASS-PAST

'Uncomfortably for Taro, Hanako was ambushing him on his way.'

(45) Taro-wa, Hanako-ni shin-are-ta.

Taro-TOP Hanako-by die-PASS-PAST

'Taro was adversely affected by Hanako dying.'/'Unfortunately for Taro, Hanako died.'

In (42), where Taro is directly affected by the action of the verb meaning 'bring up', we have an objective description of an event. It can be used to describe a situation in which Taro is adversely affected, but this is not necessarily implied. If his daughter was brought up as described in (43), that is, if Taro is indirectly affected by the action, the adversative reading arises. In some contexts, however, a neutral reading is possible. In (44) we see no direct relation between Taro and Hanako or the path that Taro is walking along. In this case, the maleficiary meaning is obvious. In the case of the indirect passive with an unaccusative verb shinu 'die' in (45), we have also the same maleficiary meaning, since Taro was adversely affected by Hanako's dying.

As we have explained in (35), we get the specific function of the Japanese passive in an effort to recover the gap between the two arguments, the passive of suffering.

Above, we presented a contrastive analysis of the Japanese and German passive constructions, focusing on their rather peripheral usages. Now we would like to come back to the passive definition we gave in (1) as a working hypothesis and reexamine it.

From the discussion so far, the working definition of the passive turns out to be inapplicable to the Japanese indirect passive. From this insufficiency of the above definition, we could argue that the passive is a language-specific category and "passive" forms of different languages can be compared only within a certain closely linked group like European languages. The "exotic" usages of the passive like the Japanese indirect passive would be a passive-like construction at best.

Or, this is as we would argue, the "traditional" definition can be modified to a more universal voice category which is compatible with a higher degree of variation among languages. Shibatani (this volume) takes this road, and proposes an alternative definition of voice.

(46) Major voice parameters (cf. Shibatani this volume)

I. The origin of an action

(a) How is the action brought about?

(b) Where does the action originate?

(c) What is the nature of the agent?

II. The development of an action

How does the action develop--beyond the agent's sphere or confined to it?

III. The termination of action

(a) Where does the action terminate?

(b) Does the action develop to its full extent and affect the patient, or does it fail to do so?

(c) Does the action develop further than its normal course such that the effect is registered in an entity other than the direct participants of the event?

Our observation so far supports this understanding of the passive voice in two ways:

1. The passive is a category which deals with transitivity: the human languages use different voice forms depending on how we perceive an event. Both in Japanese and in German, we observed that the prototypical usage of the passive construction respectively made by a prototypical transitive verb and the deviance from the prototype (intransitive passive or indirect passive in Japanese) result in a language-specific usage of the passive form.

2. Another parameter of the voice choice is the point of view, that is, from where the described action is observed. The Japanese indirect passive is a choice if one perceives an event from the perspective of the experiencer, while the prototypical passives in German and Japanese, ones with a typical transitive verb, profile the perspective from the patient.

Received 20 January 2004 Revised version received 20 September 2005


Eroms, Hans-Werner (1990). Zur Entwicklung der Passivperiphrasen im Deutschen. In Akten der Tagung 'Neuere Methoden in der historischen Syntax des Deutschen" Eichstatt 1989, A. Betten (ed.), 82-97. Tubingen: Niemeyer.

--(2000). Syntax der deutschen Sprache. Berlin and New York: Walter de Gruyter.

Felix, W. Sascha (2003). Theta parameterization: predicate-argument structures in Japanese. Linguistische Berichte 194, 131-152.

Hashimoto, Shinkichi (1969). Zyosi, zyodosi no kenkyu [Studies of Particles and Auxiliary Verbs]. Tokyo: Iwanami.

Hopper, Paul J.; and Thompson, Sandra A. (1980). Transitivity in grammar and discourse. Language 56, 251-299.

Karashima, Mie (1993). Ru/raru no sonkeyoho no hassei to tenkai: komonjo hoka no yorei kara [Derivation and development of the honorific uses of -ru and -raru: based on examples taken from the komonjo and other ancient texts]. Kokugogaku 172, 1-14.

Kotin, Michail (1995). Das Deutsche als "werden"-Sprache. (Synchronie und Diachronie der "werden"-Perspektive im deutschen Verbalsystem). In Das Wort, Germanistisches Jahrbuch 1995, A. Ehlert (ed.), 12-27. Bonn: DAAD.

Kuginuki, Toru (1991). Jyodoshi ru/raru, su/sasu seiritu no rekisitekijoken ni tsuite [The historical condition of the formation of auxiliary verbs ru/raru (passive) and su/sasu (causative)]. Kokugogaku 164, 15-28.

Muraki, Shinjiro (1991). Voice no Category to Bunkouzou no Level [Category of the diathesis and level of the sentence structure]. In Nihon-go no Voice to Tadousei [Diathesis and Transitivity in Japanese], Y. Nitta (ed.), 1-30. Tokyo: Kuroshio.

Nitta, Yoshio (1991). Voice teki hyougen to jikoseigyosei. [Expressions of voice and self-control]. In Nihon-go no Voice to Tadousei [Diathesis and Transitivity in Japanese], Y. Nitta (ed.), 31-57. Tokyo: Kuroshio.

Shibatani, Masayoshi (2000). Voice. In Bun no kokkaku [Architecture of Sentences], Nihongo no bunpo [Japanese Syntax], Vol. 1, Y. Nitta and T. Masuoka (eds.) 117-186. Tokyo: Iwanami.

Silverstein, Michael (1976). Hierarchy of features and ergativity. In Grammatical Categories in Australian Languages, R. M. W. Dixon (ed.), 112-171. Canberra: Humanities Press.

Tanaka, Shin (2003). Im Anfang ist das Topik--Topik-Merge: Generierung des NichtArgument-Topiks. Neue Beitrage zur Germanistik 2(2), 48-60.

Tsunoda, Tasaku (1991). Sekai-no-gengo to Nihon-go [Languages in the World and Japanese]. Tokyo: Kuroshio.

Weinrich, Harald (1993). Textgrammatik der deutschen Sprache. Mannheim: Dudenverlag.

Zifonun, Gisela; Hoffman, Ludger; and Strecker, Bruno (1997). Grammatik der deutschen Sprache. 3 vols. Berlin and New York. Walter de Gruyter.


(1.) This research is supported by the Grant-in-Aid for Scientific Research, Grant No. 15520243, category C (2) for Tomoaki Seino. Correspondence address: Tomoaki Seino, Chiba University, Inageku, Yayoicho 1-33, 263-8522 Chiba, Japan. E-mail: seinotom@

(2.) The passive auxiliary has two morphological realisations: rareru is added to the stem of vowel-stem verbs, and areru to the stem of the consonant-stem verbs. For example, home-ru 'commend'--home-rareru 'be commended' and koros-u 'kill'--koros-areru 'be killed'.

(3.) Here we use the terminology "intransitive passive" for the type of passive that has no accusative object to promote. Though this type of passive, which is often called the impersonal passive, can be built with a transitive verb, we choose the terminology "intransitive passive" to make the relevance of the transitivity for this type of passive clear.

(4.) An assertive illocution can be regarded as unmarked, while interrogative or imperative illocutions are marked, for which there are two reasons. The most standard reason is that the assertive sentence usually has an unmarked form, while the interrogative or imperative has a marked form. The other, more specific reason comes from the transitive parameter of Hopper and Thompson (1980) (our Table 1): the parameter of mode says that an action which either did not occur or is occurring in a contingent world is less transitive, that is, less unmarked in the discourse in which the transitivity is highlighted.

(5.) Otherwise, the ga-phrase must be accented (indicated by italicization in [ii]) and receives a contrastive interpretation.

(i) Kare-wa kanojo-nimusuko-wo nagur-are-ta.

he-TOP she-by son-acc hit-PASS-PAST

'He was adversely affected because his son was beaten.'

(ii) Kare-ga kanojo-ni musuko-wo nagur-are-ta.

he-NOM she-by son-ACC hit-PASS-PAST

'He is the person who was adversely affected because his son was beaten.'

(6.) For the exact sense in which Hole's binding mechanism is special, cf. his contribution to this volume.


Chiba University
Table 1. Parameters of transitivity (Hopper and Thompson 1980)

 High Low

A. Participants 2 or more participants, 1 participant
 A and O
B. Kinesis action non-action
C. Aspect telic atelic
D. Punctuality punctual non-punctual
E. Volitionality volitional non-volitional
F. Affirmation affirmative negative
G. Mode realis irrealis
H. Agency A high in potency A low in potency
I. Affectedness of O O totally affected O not affected
J. Individuation O highly individuated O non-individuated

Table 2. Passive affinity

Transitivity Verbs Active Passive %

transitive high ermorden 'murder' 35 24 40.1
 toten 'kill' 120 69 36.5
 stehlen 'steal' 58 31 34.8
 brechen 'break' 122 15 10.9
 finden "find' 2247 153 6.4
 schlagen 'hit' 658 36 5.2
 essen 'eat' 252 3 1.2
 lesen 'read' 462 7 1.5
 empfehlen 'recommend' 142 11 7.2
 wunschen 'wish' 167 4 2.3
 feiern 'celebrate' 144 21 12.7
intransitive arbeiten 'work' 210 14 6.3
 weinen 'cry' 225 3 1.3
 low bleiben 'stay' 2511 1 0

Table 3. Iterativity/habituality

 toten schlagen feiern arbeiten
 % % % %

Iterative 4.4 5.3 21.4 62.5
Semelfactive 95.6 94.7 78.6 37.5
COPYRIGHT 2006 Walter de Gruyter GmbH & Co. KG
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2006 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Seino, Tomoaki; Tanaka, Shin
Publication:Linguistics: an interdisciplinary journal of the language sciences
Geographic Code:9JAPA
Date:Mar 1, 2006
Previous Article:Transitivity pairs, markedness, and diachronic stability *.
Next Article:Rethinking the relationship between SELF-intensifiers and reflexives.

Related Articles
A complementation approach to Chinese passives and its consequences(*).
Reflexives and resultatives: some differences between English and German *.
Topic-worthiness in German and English *.
Introduction: operations on argument structure *.
4. Pragmatically motivated voice systems.
Extra argumentality--affectees, landmarks, and voice *.
Lessons learned from 17 years of HIV-prevention research.
Don't forget the need for back-up to wind turbines.
Minimalist essays.
Feature-based argument mapping and animacy optimization in impersonal passives.

Terms of use | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters